An End-Use Integrated Agent-Based Model to Simulate Consumer Demand for a Water Utility

Accurately estimating residential water use is becoming an increasingly important issue to the water utilities due to limited supply, increased demand caused by a growing population and urbanization, and seasonal variation of global climatic conditions. Most existing demand estimation methods are deterministic or stochastic. Deterministic methods are based on end-use models, which evaluate household water use at the appliance level. Stochastic models predict water use using empirical relationships based on predictors such as population size, water pricing, rainfall, and daily maximum temperature. However, consumer demand varies due to factors such as water pricing, demographics, weather, technology, consumer habits, and local conservation and restriction programs, which have not been combined in any single application before. In this research a new modeling approach is developed for estimating residential monthly water use by integrating a deterministic end-use model within an agent-based modeling framework to model the combined effect of all demand variables. Heterogeneity in consumer behavior is incorporated through appliance type and lot size information; stochasticity in demand is included through climate data and population growth. The proposed model is applied to a Texas utility, and analysis of the results shows that the proposed method could simulate residential demand with reasonable accuracy. This approach can be a useful demand management tool for water utilities.